parsing_cocci/lexer_cocci.mll

   1 (*
   2 * Copyright 2005-2009, Ecole des Mines de Nantes, University of Copenhagen
   3 * Yoann Padioleau, Julia Lawall, Rene Rydhof Hansen, Henrik Stuart, Gilles Muller
   4 * This file is part of Coccinelle.
   5 *
   6 * Coccinelle is free software: you can redistribute it and/or modify
   7 * it under the terms of the GNU General Public License as published by
   8 * the Free Software Foundation, according to version 2 of the License.
   9 *
  10 * Coccinelle is distributed in the hope that it will be useful,
  11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 * GNU General Public License for more details.
  14 *
  15 * You should have received a copy of the GNU General Public License
  16 * along with Coccinelle.  If not, see <http://www.gnu.org/licenses/>.
  17 *
  18 * The authors reserve the right to distribute this or future versions of
  19 * Coccinelle under other licenses.
  20 *)
  21
  22
  23 {
  24 open Parser_cocci_menhir
  25 module D = Data
  26 module Ast = Ast_cocci
  27 module Ast0 = Ast0_cocci
  28 module P = Parse_aux
  29 exception Lexical of string
  30 let tok = Lexing.lexeme
  31
  32 let line = ref 1
  33 let logical_line = ref 0
  34
  35 (* ---------------------------------------------------------------------- *)
  36 (* control codes *)
  37
  38 (* Defined in data.ml
  39 type line_type = MINUS | OPTMINUS | UNIQUEMINUS | PLUS | CONTEXT | UNIQUE | OPT
  40 *)
  41
  42 let current_line_type = ref (D.CONTEXT,!line,!logical_line)
  43
  44 let prev_plus = ref false
  45 let line_start = ref 0 (* offset of the beginning of the line *)
  46 let get_current_line_type lexbuf =
  47   let (c,l,ll) = !current_line_type in
  48   let lex_start = Lexing.lexeme_start lexbuf in
  49   let preceeding_spaces =
  50     if !line_start < 0 then 0 else lex_start - !line_start in
  51   line_start := -1;
  52   prev_plus := (c = D.PLUS);
  53   (c,l,ll,lex_start,preceeding_spaces,[],[],Ast0.NoMetaPos)
  54 let current_line_started = ref false
  55 let col_zero = ref true
  56
  57 let reset_line lexbuf =
  58   line := !line + 1;
  59   current_line_type := (D.CONTEXT,!line,!logical_line);
  60   current_line_started := false;
  61   col_zero := true;
  62   line_start := Lexing.lexeme_start lexbuf + 1
  63
  64 let started_line = ref (-1)
  65
  66 let start_line seen_char =
  67   current_line_started := true;
  68   col_zero := false;
  69   (if seen_char && not(!line = !started_line)
  70   then
  71     begin
  72       started_line := !line;
  73       logical_line := !logical_line + 1
  74     end)
  75
  76 let pass_zero _ = col_zero := false
  77
  78 let lexerr s1 s2 = raise (Lexical (Printf.sprintf "%s%s" s1 s2))
  79
  80 let add_current_line_type x =
  81   match (x,!current_line_type) with
  82     (D.MINUS,(D.CONTEXT,ln,lln))  ->
  83       current_line_type := (D.MINUS,ln,lln)
  84   | (D.MINUS,(D.UNIQUE,ln,lln))   ->
  85       current_line_type := (D.UNIQUEMINUS,ln,lln)
  86   | (D.MINUS,(D.OPT,ln,lln))      ->
  87       current_line_type := (D.OPTMINUS,ln,lln)
  88   | (D.PLUS,(D.CONTEXT,ln,lln))   ->
  89       current_line_type := (D.PLUS,ln,lln)
  90   | (D.UNIQUE,(D.CONTEXT,ln,lln)) ->
  91       current_line_type := (D.UNIQUE,ln,lln)
  92   | (D.OPT,(D.CONTEXT,ln,lln))    ->
  93       current_line_type := (D.OPT,ln,lln)
  94   | _ -> lexerr "invalid control character combination" ""
  95
  96 let check_minus_context_linetype s =
  97   match !current_line_type with
  98     (D.PLUS,_,_) -> lexerr "invalid in a + context: " s
  99   | _ -> ()
 100
 101 let check_context_linetype s =
 102   match !current_line_type with
 103     (D.CONTEXT,_,_) -> ()
 104   | _ -> lexerr "invalid in a nonempty context: " s
 105
 106 let check_plus_linetype s =
 107   match !current_line_type with
 108     (D.PLUS,_,_) -> ()
 109   | _ -> lexerr "invalid in a non + context: " s
 110
 111 let check_arity_context_linetype s =
 112   match !current_line_type with
 113     (D.CONTEXT,_,_) | (D.PLUS,_,_) | (D.UNIQUE,_,_) | (D.OPT,_,_) -> ()
 114   | _ -> lexerr "invalid in a nonempty context: " s
 115
 116 let process_include start finish str =
 117   (match !current_line_type with
 118     (D.PLUS,_,_) ->
 119       (try
 120         let _ = Str.search_forward (Str.regexp "\\.\\.\\.") str start in
 121         lexerr "... not allowed in + include" ""
 122       with Not_found -> ())
 123   | _ -> ());
 124   String.sub str (start + 1) (finish - start - 1)
 125
 126 (* ---------------------------------------------------------------------- *)
 127 type pm = PATCH | MATCH | UNKNOWN
 128
 129 let pm = ref UNKNOWN
 130
 131 let patch_or_match = function
 132     PATCH ->
 133       (match !pm with
 134         MATCH -> lexerr "- or + not allowed in the first column for a match" ""
 135       | PATCH -> ()
 136       | UNKNOWN -> Flag.sgrep_mode2 := false; pm := PATCH)
 137   | MATCH ->
 138       (match !pm with
 139         PATCH -> lexerr "* not allowed in the first column for a patch" ""
 140       | MATCH -> ()
 141       | UNKNOWN -> Flag.sgrep_mode2 := true; pm := MATCH)
 142   | _ -> failwith "unexpected argument"
 143
 144 (* ---------------------------------------------------------------------- *)
 145 (* identifiers, including metavariables *)
 146
 147 let metavariables = (Hashtbl.create(100) : (string, D.clt -> token) Hashtbl.t)
 148
 149 let all_metavariables =
 150   (Hashtbl.create(100) : (string,(string * (D.clt -> token)) list) Hashtbl.t)
 151
 152 let type_names = (Hashtbl.create(100) : (string, D.clt -> token) Hashtbl.t)
 153
 154 let declarer_names = (Hashtbl.create(100) : (string, D.clt -> token) Hashtbl.t)
 155
 156 let iterator_names = (Hashtbl.create(100) : (string, D.clt -> token) Hashtbl.t)
 157
 158 let rule_names = (Hashtbl.create(100) : (string, unit) Hashtbl.t)
 159
 160 let check_var s linetype =
 161   let fail _ =
 162     if (!Data.in_prolog || !Data.in_rule_name) &&
 163       Str.string_match (Str.regexp "<.*>") s 0
 164     then TPathIsoFile s
 165     else
 166       try (Hashtbl.find metavariables s) linetype
 167       with Not_found ->
 168         (try (Hashtbl.find type_names s) linetype
 169         with Not_found ->
 170           (try (Hashtbl.find declarer_names s) linetype
 171           with Not_found ->
 172             (try (Hashtbl.find iterator_names s) linetype
 173             with Not_found -> TIdent (s,linetype)))) in
 174   if !Data.in_meta or !Data.in_rule_name
 175   then (try Hashtbl.find rule_names s; TRuleName s with Not_found -> fail())
 176   else fail()
 177
 178 let id_tokens lexbuf =
 179   let s = tok lexbuf in
 180   let linetype = get_current_line_type lexbuf in
 181   let in_rule_name = !Data.in_rule_name in
 182   let in_meta = !Data.in_meta in
 183   let in_iso = !Data.in_iso in
 184   let in_prolog = !Data.in_prolog in
 185   match s with
 186     "identifier" when in_meta -> check_arity_context_linetype s; TIdentifier
 187   | "type" when in_meta ->       check_arity_context_linetype s; TType
 188   | "parameter" when in_meta ->  check_arity_context_linetype s; TParameter
 189   | "constant"  when in_meta ->  check_arity_context_linetype s; TConstant
 190   | "generated" when in_rule_name && not (!Flag.make_hrule = None) ->
 191       check_arity_context_linetype s; TGenerated
 192   | "expression" when in_meta || in_rule_name ->
 193       check_arity_context_linetype s; TExpression
 194   | "initialiser" when in_meta || in_rule_name ->
 195       check_arity_context_linetype s; TInitialiser
 196   | "initializer" when in_meta || in_rule_name ->
 197       check_arity_context_linetype s; TInitialiser
 198   | "idexpression" when in_meta ->
 199       check_arity_context_linetype s; TIdExpression
 200   | "statement" when in_meta ->  check_arity_context_linetype s; TStatement
 201   | "function"  when in_meta ->  check_arity_context_linetype s; TFunction
 202   | "local" when in_meta ->      check_arity_context_linetype s; TLocal
 203   | "list" when in_meta ->       check_arity_context_linetype s; Tlist
 204   | "fresh" when in_meta ->      check_arity_context_linetype s; TFresh
 205   | "typedef" when in_meta ->    check_arity_context_linetype s; TTypedef
 206   | "declarer" when in_meta ->   check_arity_context_linetype s; TDeclarer
 207   | "iterator" when in_meta ->   check_arity_context_linetype s; TIterator
 208   | "name" when in_meta ->       check_arity_context_linetype s; TName
 209   | "position" when in_meta ->   check_arity_context_linetype s; TPosition
 210   | "any" when in_meta ->        check_arity_context_linetype s; TPosAny
 211   | "pure" when in_meta && in_iso ->
 212       check_arity_context_linetype s; TPure
 213   | "context" when in_meta && in_iso ->
 214       check_arity_context_linetype s; TContext
 215   | "error" when in_meta ->      check_arity_context_linetype s; TError
 216   | "words" when in_meta ->      check_context_linetype s; TWords
 217
 218   | "using" when in_rule_name || in_prolog ->  check_context_linetype s; TUsing
 219   | "disable" when in_rule_name ->  check_context_linetype s; TDisable
 220   | "extends" when in_rule_name -> check_context_linetype s; TExtends
 221   | "depends" when in_rule_name -> check_context_linetype s; TDepends
 222   | "on" when in_rule_name      -> check_context_linetype s; TOn
 223   | "ever" when in_rule_name    -> check_context_linetype s; TEver
 224   | "never" when in_rule_name   -> check_context_linetype s; TNever
 225   | "exists" when in_rule_name  -> check_context_linetype s; TExists
 226   | "forall" when in_rule_name  -> check_context_linetype s; TForall
 227   | "reverse" when in_rule_name -> check_context_linetype s; TReverse
 228   | "script" when in_rule_name  -> check_context_linetype s; TScript
 229   | "initialize" when in_rule_name -> check_context_linetype s; TInitialize
 230   | "finalize" when in_rule_name   -> check_context_linetype s; TFinalize
 231
 232   | "char" ->       Tchar     linetype
 233   | "short" ->      Tshort    linetype
 234   | "int" ->        Tint      linetype
 235   | "double" ->     Tdouble   linetype
 236   | "float" ->      Tfloat    linetype
 237   | "long" ->       Tlong     linetype
 238   | "void" ->       Tvoid     linetype
 239   | "struct" ->     Tstruct   linetype
 240   | "union" ->      Tunion    linetype
 241   | "enum" ->       Tenum     linetype
 242   | "unsigned" ->   Tunsigned linetype
 243   | "signed" ->     Tsigned   linetype
 244
 245   | "auto"  ->      Tauto     linetype
 246   | "register" ->   Tregister linetype
 247   | "extern" ->     Textern   linetype
 248   | "static" ->     Tstatic   linetype
 249   | "inline" ->     Tinline   linetype
 250   | "typedef" ->    Ttypedef  linetype
 251
 252   | "const" ->      Tconst    linetype
 253   | "volatile" ->   Tvolatile linetype
 254
 255   | "if" ->         TIf       linetype
 256   | "else" ->       TElse     linetype
 257   | "while" ->      TWhile    linetype
 258   | "do" ->         TDo       linetype
 259   | "for" ->        TFor      linetype
 260   | "switch" ->     TSwitch   linetype
 261   | "case" ->       TCase     linetype
 262   | "default" ->    TDefault  linetype
 263   | "return" ->     TReturn   linetype
 264   | "break" ->      TBreak    linetype
 265   | "continue" ->   TContinue linetype
 266   | "goto" ->       TGoto     linetype
 267
 268   | "sizeof" ->     TSizeof   linetype
 269
 270   | "Expression"     -> TIsoExpression
 271   | "ArgExpression"  -> TIsoArgExpression
 272   | "TestExpression" -> TIsoTestExpression
 273   | "Statement"      -> TIsoStatement
 274   | "Declaration"    -> TIsoDeclaration
 275   | "Type"           -> TIsoType
 276   | "TopLevel"       -> TIsoTopLevel
 277
 278   | s -> check_var s linetype
 279
 280 let mkassign op lexbuf =
 281   TAssign (Ast.OpAssign op, (get_current_line_type lexbuf))
 282
 283 let init _ =
 284   line := 1;
 285   logical_line := 0;
 286   prev_plus := false;
 287   line_start := 0;
 288   current_line_started := false;
 289   col_zero := true;
 290   pm := UNKNOWN;
 291   Data.in_rule_name := false;
 292   Data.in_meta := false;
 293   Data.in_prolog := false;
 294   Data.inheritable_positions := [];
 295   Hashtbl.clear all_metavariables;
 296   Hashtbl.clear Data.all_metadecls;
 297   Hashtbl.clear metavariables;
 298   Hashtbl.clear type_names;
 299   Hashtbl.clear rule_names;
 300   let get_name (_,x) = x in
 301   Data.add_id_meta :=
 302     (fun name constraints pure ->
 303       let fn clt = TMetaId(name,constraints,pure,clt) in
 304       Hashtbl.replace metavariables (get_name name) fn);
 305   Data.add_fresh_id_meta :=
 306     (fun name ->
 307       let fn clt = TMetaId(name,[],Ast0.Impure,clt) in
 308       Hashtbl.replace metavariables (get_name name) fn);
 309   Data.add_type_meta :=
 310     (fun name pure ->
 311       let fn clt = TMetaType(name,pure,clt) in
 312       Hashtbl.replace metavariables (get_name name) fn);
 313   Data.add_init_meta :=
 314     (fun name pure ->
 315       let fn clt = TMetaInit(name,pure,clt) in
 316       Hashtbl.replace metavariables (get_name name) fn);
 317   Data.add_param_meta :=
 318     (function name -> function pure ->
 319       let fn clt = TMetaParam(name,pure,clt) in
 320       Hashtbl.replace metavariables (get_name name) fn);
 321   Data.add_paramlist_meta :=
 322     (function name -> function lenname -> function pure ->
 323       let fn clt = TMetaParamList(name,lenname,pure,clt) in
 324       Hashtbl.replace metavariables (get_name name) fn);
 325   Data.add_const_meta :=
 326     (fun tyopt name constraints pure ->
 327       let fn clt = TMetaConst(name,constraints,pure,tyopt,clt) in
 328       Hashtbl.replace metavariables (get_name name) fn);
 329   Data.add_err_meta :=
 330     (fun name constraints pure ->
 331       let fn clt = TMetaErr(name,constraints,pure,clt) in
 332       Hashtbl.replace metavariables (get_name name) fn);
 333   Data.add_exp_meta :=
 334     (fun tyopt name constraints pure ->
 335       let fn clt = TMetaExp(name,constraints,pure,tyopt,clt) in
 336       Hashtbl.replace metavariables (get_name name) fn);
 337   Data.add_idexp_meta :=
 338     (fun tyopt name constraints pure ->
 339       let fn clt = TMetaIdExp(name,constraints,pure,tyopt,clt) in
 340       Hashtbl.replace metavariables (get_name name) fn);
 341   Data.add_local_idexp_meta :=
 342     (fun tyopt name constraints pure ->
 343       let fn clt = TMetaLocalIdExp(name,constraints,pure,tyopt,clt) in
 344       Hashtbl.replace metavariables (get_name name) fn);
 345   Data.add_explist_meta :=
 346     (function name -> function lenname -> function pure ->
 347       let fn clt = TMetaExpList(name,lenname,pure,clt) in
 348       Hashtbl.replace metavariables (get_name name) fn);
 349   Data.add_stm_meta :=
 350     (function name -> function pure ->
 351       let fn clt = TMetaStm(name,pure,clt) in
 352       Hashtbl.replace metavariables (get_name name) fn);
 353   Data.add_stmlist_meta :=
 354     (function name -> function pure ->
 355       let fn clt = TMetaStmList(name,pure,clt) in
 356       Hashtbl.replace metavariables (get_name name) fn);
 357   Data.add_func_meta :=
 358     (fun name constraints pure ->
 359       let fn clt = TMetaFunc(name,constraints,pure,clt) in
 360       Hashtbl.replace metavariables (get_name name) fn);
 361   Data.add_local_func_meta :=
 362     (fun name constraints pure ->
 363       let fn clt = TMetaLocalFunc(name,constraints,pure,clt) in
 364       Hashtbl.replace metavariables (get_name name) fn);
 365   Data.add_iterator_meta :=
 366     (fun name constraints pure ->
 367       let fn clt = TMetaIterator(name,constraints,pure,clt) in
 368       Hashtbl.replace metavariables (get_name name) fn);
 369   Data.add_declarer_meta :=
 370     (fun name constraints pure ->
 371       let fn clt = TMetaDeclarer(name,constraints,pure,clt) in
 372       Hashtbl.replace metavariables (get_name name) fn);
 373   Data.add_pos_meta :=
 374     (fun name constraints any ->
 375       let fn ((d,ln,_,_,_,_,_,_) as clt) =
 376         (if d = Data.PLUS
 377         then
 378           failwith
 379             (Printf.sprintf "%d: positions only allowed in minus code" ln));
 380         TMetaPos(name,constraints,any,clt) in
 381       Hashtbl.replace metavariables (get_name name) fn);
 382   Data.add_type_name :=
 383     (function name ->
 384       let fn clt = TTypeId(name,clt) in
 385       Hashtbl.replace type_names name fn);
 386   Data.add_declarer_name :=
 387     (function name ->
 388       let fn clt = TDeclarerId(name,clt) in
 389       Hashtbl.replace declarer_names name fn);
 390   Data.add_iterator_name :=
 391     (function name ->
 392       let fn clt = TIteratorId(name,clt) in
 393       Hashtbl.replace iterator_names name fn);
 394   Data.init_rule := (function _ -> Hashtbl.clear metavariables);
 395   Data.install_bindings :=
 396     (function parent ->
 397       List.iter (function (name,fn) -> Hashtbl.add metavariables name fn)
 398         (Hashtbl.find all_metavariables parent))
 399
 400 let drop_spaces s =
 401   let len = String.length s in
 402   let rec loop n =
 403     if n = len
 404     then n
 405     else
 406       if List.mem (String.get s n) [' ';'\t']
 407       then loop (n+1)
 408       else n in
 409   let start = loop 0 in
 410   String.sub s start (len - start)
 411 }
 412
 413 (* ---------------------------------------------------------------------- *)
 414 (* tokens *)
 415
 416 let letter = ['A'-'Z' 'a'-'z' '_']
 417 let digit  = ['0'-'9']
 418
 419 let dec = ['0'-'9']
 420 let oct = ['0'-'7']
 421 let hex = ['0'-'9' 'a'-'f' 'A'-'F']
 422
 423 let decimal = ('0' | (['1'-'9'] dec*))
 424 let octal   = ['0']        oct+
 425 let hexa    = ("0x" |"0X") hex+
 426
 427 let pent   = dec+
 428 let pfract = dec+
 429 let sign = ['-' '+']
 430 let exp  = ['e''E'] sign? dec+
 431 let real = pent exp | ((pent? '.' pfract | pent '.' pfract? ) exp?)
 432
 433
 434 rule token = parse
 435   | [' ' '\t'  ]+             { start_line false; token lexbuf }
 436   | ['\n' '\r' '\011' '\012'] { reset_line lexbuf; token lexbuf }
 437
 438   | "//" [^ '\n']* { start_line false; token lexbuf }
 439
 440   | "@@" { start_line true; TArobArob }
 441   | "@"  { pass_zero();
 442            if !Data.in_rule_name or not !current_line_started
 443            then (start_line true; TArob)
 444            else (check_minus_context_linetype "@"; TPArob) }
 445
 446   | "WHEN" | "when"
 447       { start_line true; check_minus_context_linetype (tok lexbuf);
 448         TWhen (get_current_line_type lexbuf) }
 449
 450   | "..."
 451       { start_line true; check_minus_context_linetype (tok lexbuf);
 452         TEllipsis (get_current_line_type lexbuf) }
 453 (*
 454   | "ooo"
 455       { start_line true; check_minus_context_linetype (tok lexbuf);
 456         TCircles (get_current_line_type lexbuf) }
 457
 458   | "***"
 459       { start_line true; check_minus_context_linetype (tok lexbuf);
 460         TStars (get_current_line_type lexbuf) }
 461 *)
 462   | "<..." { start_line true; check_context_linetype (tok lexbuf);
 463              TOEllipsis (get_current_line_type lexbuf) }
 464   | "...>" { start_line true; check_context_linetype (tok lexbuf);
 465              TCEllipsis (get_current_line_type lexbuf) }
 466   | "<+..." { start_line true; check_context_linetype (tok lexbuf);
 467              TPOEllipsis (get_current_line_type lexbuf) }
 468   | "...+>" { start_line true; check_context_linetype (tok lexbuf);
 469              TPCEllipsis (get_current_line_type lexbuf) }
 470 (*
 471   | "<ooo" { start_line true; check_context_linetype (tok lexbuf);
 472              TOCircles (get_current_line_type lexbuf) }
 473   | "ooo>" { start_line true; check_context_linetype (tok lexbuf);
 474              TCCircles (get_current_line_type lexbuf) }
 475
 476   | "<***" { start_line true; check_context_linetype (tok lexbuf);
 477              TOStars (get_current_line_type lexbuf) }
 478   | "***>" { start_line true; check_context_linetype (tok lexbuf);
 479              TCStars (get_current_line_type lexbuf) }
 480 *)
 481   | "-" { pass_zero();
 482           if !current_line_started
 483           then (start_line true; TMinus (get_current_line_type lexbuf))
 484           else (patch_or_match PATCH;
 485                 add_current_line_type D.MINUS; token lexbuf) }
 486   | "+" { pass_zero();
 487           if !current_line_started
 488           then (start_line true; TPlus (get_current_line_type lexbuf))
 489           else if !Data.in_meta
 490           then TPlus0
 491           else (patch_or_match PATCH;
 492                 add_current_line_type D.PLUS; token lexbuf) }
 493   | "?" { pass_zero();
 494           if !current_line_started
 495           then (start_line true; TWhy (get_current_line_type lexbuf))
 496           else if !Data.in_meta
 497           then TWhy0
 498           else (add_current_line_type D.OPT; token lexbuf) }
 499   | "!" { pass_zero();
 500           if !current_line_started
 501           then (start_line true; TBang (get_current_line_type lexbuf))
 502           else if !Data.in_meta
 503           then TBang0
 504           else (add_current_line_type D.UNIQUE; token lexbuf) }
 505   | "(" { if not !col_zero
 506           then (start_line true; TOPar (get_current_line_type lexbuf))
 507           else
 508             (start_line true; check_context_linetype (tok lexbuf);
 509              TOPar0 (get_current_line_type lexbuf))}
 510   | "\\(" { start_line true; TOPar0 (get_current_line_type lexbuf) }
 511   | "|" { if not (!col_zero)
 512           then (start_line true; TOr(get_current_line_type lexbuf))
 513           else (start_line true;
 514                 check_context_linetype (tok lexbuf);
 515                 TMid0 (get_current_line_type lexbuf))}
 516   | "\\|" { start_line true; TMid0 (get_current_line_type lexbuf) }
 517   | ")" { if not !col_zero
 518           then (start_line true; TCPar (get_current_line_type lexbuf))
 519           else
 520             (start_line true; check_context_linetype (tok lexbuf);
 521              TCPar0 (get_current_line_type lexbuf))}
 522   | "\\)" { start_line true; TCPar0 (get_current_line_type lexbuf) }
 523
 524   | '[' { start_line true; TOCro (get_current_line_type lexbuf)   }
 525   | ']' { start_line true; TCCro (get_current_line_type lexbuf)   }
 526   | '{' { start_line true; TOBrace (get_current_line_type lexbuf) }
 527   | '}' { start_line true; TCBrace (get_current_line_type lexbuf) }
 528
 529   | "->"           { start_line true; TPtrOp (get_current_line_type lexbuf)  }
 530   | '.'            { start_line true; TDot (get_current_line_type lexbuf)    }
 531   | ','            { start_line true; TComma (get_current_line_type lexbuf)  }
 532   | ";"            { start_line true;
 533                      if !Data.in_meta
 534                      then TMPtVirg (* works better with tokens_all *)
 535                      else TPtVirg (get_current_line_type lexbuf) }
 536
 537
 538   | '*'            { pass_zero();
 539                      if !current_line_started
 540                      then
 541                        (start_line true; TMul (get_current_line_type lexbuf))
 542                      else
 543                        (patch_or_match MATCH;
 544                         add_current_line_type D.MINUS; token lexbuf) }
 545   | '/'            { start_line true;
 546                      TDmOp (Ast.Div,get_current_line_type lexbuf) }
 547   | '%'            { start_line true;
 548                      TDmOp (Ast.Mod,get_current_line_type lexbuf) }
 549   | '~'            { start_line true;  TTilde (get_current_line_type lexbuf) }
 550
 551   | "++"           { start_line true;  TInc (get_current_line_type lexbuf) }
 552   | "--"           { start_line true;  TDec (get_current_line_type lexbuf) }
 553
 554   | "="            { start_line true; TEq (get_current_line_type lexbuf) }
 555
 556   | "-="           { start_line true; mkassign Ast.Minus lexbuf }
 557   | "+="           { start_line true; mkassign Ast.Plus lexbuf }
 558
 559   | "*="           { start_line true; mkassign Ast.Mul lexbuf }
 560   | "/="           { start_line true; mkassign Ast.Div lexbuf }
 561   | "%="           { start_line true; mkassign Ast.Mod lexbuf }
 562
 563   | "&="           { start_line true; mkassign Ast.And lexbuf }
 564   | "|="           { start_line true; mkassign Ast.Or lexbuf }
 565   | "^="           { start_line true; mkassign Ast.Xor lexbuf }
 566
 567   | "<<="          { start_line true; mkassign Ast.DecLeft lexbuf }
 568   | ">>="          { start_line true; mkassign Ast.DecRight lexbuf }
 569
 570   | ":"            { start_line true; TDotDot (get_current_line_type lexbuf) }
 571
 572   | "=="           { start_line true; TEqEq   (get_current_line_type lexbuf) }
 573   | "!="           { start_line true; TNotEq  (get_current_line_type lexbuf) }
 574   | ">="           { start_line true;
 575                      TLogOp(Ast.SupEq,get_current_line_type lexbuf) }
 576   | "<="           { start_line true;
 577                      TLogOp(Ast.InfEq,get_current_line_type lexbuf) }
 578   | "<"            { start_line true;
 579                      TLogOp(Ast.Inf,get_current_line_type lexbuf) }
 580   | ">"            { start_line true;
 581                      TLogOp(Ast.Sup,get_current_line_type lexbuf) }
 582
 583   | "&&"           { start_line true; TAndLog (get_current_line_type lexbuf) }
 584   | "||"           { start_line true; TOrLog  (get_current_line_type lexbuf) }
 585
 586   | ">>"           { start_line true;
 587                      TShOp(Ast.DecRight,get_current_line_type lexbuf) }
 588   | "<<"           { start_line true;
 589                      TShOp(Ast.DecLeft,get_current_line_type lexbuf) }
 590
 591   | "&"            { start_line true; TAnd    (get_current_line_type lexbuf) }
 592   | "^"            { start_line true; TXor(get_current_line_type lexbuf) }
 593
 594   | ( ("#" [' ' '\t']*  "define" [' ' '\t']+))
 595     ( (letter (letter |digit)*) as ident)
 596       { start_line true;
 597         let (arity,line,lline,offset,col,strbef,straft,pos) as lt =
 598           get_current_line_type lexbuf in
 599         let off = String.length "#define " in
 600         (* -1 in the code below because the ident is not at the line start *)
 601         TDefine
 602           (lt,
 603            check_var ident
 604              (arity,line,lline,offset+off,(-1),[],[],Ast0.NoMetaPos)) }
 605   | ( ("#" [' ' '\t']*  "define" [' ' '\t']+))
 606     ( (letter (letter | digit)*) as ident)
 607     '('
 608       { start_line true;
 609         let (arity,line,lline,offset,col,strbef,straft,pos) as lt =
 610           get_current_line_type lexbuf in
 611         let off = String.length "#define " in
 612         TDefineParam
 613         (lt,
 614          check_var ident
 615            (* why pos here but not above? *)
 616            (arity,line,lline,offset+off,(-1),strbef,straft,pos),
 617          offset + off + (String.length ident)) }
 618   | "#" [' ' '\t']* "include" [' ' '\t']* '"' [^ '"']+ '"'
 619       { TIncludeL
 620           (let str = tok lexbuf in
 621           let start = String.index str '"' in
 622           let finish = String.rindex str '"' in
 623           start_line true;
 624           (process_include start finish str,get_current_line_type lexbuf)) }
 625   | "#" [' ' '\t']* "include" [' ' '\t']* '<' [^ '>']+ '>'
 626       { TIncludeNL
 627           (let str = tok lexbuf in
 628           let start = String.index str '<' in
 629           let finish = String.rindex str '>' in
 630           start_line true;
 631           (process_include start finish str,get_current_line_type lexbuf)) }
 632   | "#" [' ' '\t']* "if" [^'\n']*
 633   | "#" [' ' '\t']* "ifdef" [^'\n']*
 634   | "#" [' ' '\t']* "ifndef" [^'\n']*
 635   | "#" [' ' '\t']* "else" [^'\n']*
 636   | "#" [' ' '\t']* "elif" [^'\n']*
 637   | "#" [' ' '\t']* "endif" [^'\n']*
 638   | "#" [' ' '\t']* "error" [^'\n']*
 639       { start_line true; check_plus_linetype (tok lexbuf);
 640         TPragma (tok lexbuf, get_current_line_type lexbuf) }
 641   | "/*"
 642       { start_line true; check_plus_linetype (tok lexbuf);
 643         (* second argument to TPragma is not quite right, because
 644            it represents only the first token of the comemnt, but that
 645            should be good enough *)
 646         TPragma ("/*"^(comment lexbuf), get_current_line_type lexbuf) }
 647   | "---" [^'\n']*
 648       { (if !current_line_started
 649       then lexerr "--- must be at the beginning of the line" "");
 650         start_line true;
 651         TMinusFile
 652           (let str = tok lexbuf in
 653           (drop_spaces(String.sub str 3 (String.length str - 3)),
 654            (get_current_line_type lexbuf))) }
 655   | "+++" [^'\n']*
 656       { (if !current_line_started
 657       then lexerr "+++ must be at the beginning of the line" "");
 658         start_line true;
 659         TPlusFile
 660           (let str = tok lexbuf in
 661           (drop_spaces(String.sub str 3 (String.length str - 3)),
 662            (get_current_line_type lexbuf))) }
 663
 664   | letter (letter | digit)*
 665       { start_line true; id_tokens lexbuf }
 666
 667   | "'" { start_line true;
 668           TChar(char lexbuf,get_current_line_type lexbuf) }
 669   | '"' { start_line true;
 670           TString(string lexbuf,(get_current_line_type lexbuf)) }
 671   | (real as x)    { start_line true;
 672                      TFloat(x,(get_current_line_type lexbuf)) }
 673   | ((( decimal | hexa | octal)
 674       ( ['u' 'U']
 675       | ['l' 'L']
 676       | (['l' 'L'] ['u' 'U'])
 677       | (['u' 'U'] ['l' 'L'])
 678       | (['u' 'U'] ['l' 'L'] ['l' 'L'])
 679       | (['l' 'L'] ['l' 'L'])
 680       )?
 681     ) as x) { start_line true; TInt(x,(get_current_line_type lexbuf)) }
 682
 683   | "<=>"          { TIso }
 684   | "=>"           { TRightIso }
 685
 686   | eof            { EOF }
 687
 688   | _ { lexerr "unrecognised symbol, in token rule: " (tok lexbuf) }
 689
 690
 691 and char = parse
 692   | (_ as x) "'"                                     { String.make 1 x }
 693   | (("\\" (oct | oct oct | oct oct oct)) as x  "'") { x }
 694   | (("\\x" (hex | hex hex)) as x  "'")       { x }
 695   | (("\\" (_ as v)) as x "'")
 696         { (match v with
 697             | 'n' -> ()  | 't' -> ()   | 'v' -> ()  | 'b' -> ()
 698             | 'r' -> ()  | 'f' -> () | 'a' -> ()
 699             | '\\' -> () | '?'  -> () | '\'' -> ()  | '"' -> ()
 700             | 'e' -> ()
 701             | _ -> lexerr "unrecognised symbol: " (tok lexbuf)
 702             );
 703           x
 704         }
 705   | _ { lexerr "unrecognised symbol: " (tok lexbuf) }
 706
 707 and string  = parse
 708   | '"'                                       { "" }
 709   | (_ as x)                   { Common.string_of_char x ^ string lexbuf }
 710   | ("\\" (oct | oct oct | oct oct oct)) as x { x ^ string lexbuf }
 711   | ("\\x" (hex | hex hex)) as x              { x ^ string lexbuf }
 712   | ("\\" (_ as v)) as x
 713        {
 714          (match v with
 715          | 'n' -> ()  | 't' -> ()   | 'v' -> ()  | 'b' -> () | 'r' -> ()
 716          | 'f' -> () | 'a' -> ()
 717          | '\\' -> () | '?'  -> () | '\'' -> ()  | '"' -> ()
 718          | 'e' -> ()
 719          | '\n' -> ()
 720          | _ -> lexerr "unrecognised symbol:" (tok lexbuf)
 721          );
 722           x ^ string lexbuf
 723        }
 724   | _ { lexerr "unrecognised symbol: " (tok lexbuf) }
 725
 726 and comment = parse
 727   | "*/"     { start_line true; tok lexbuf }
 728   | ['\n' '\r' '\011' '\012']
 729       { reset_line lexbuf; let s = tok lexbuf in s ^ comment lexbuf }
 730   | "+" { pass_zero();
 731           if !current_line_started
 732           then (start_line true; let s = tok lexbuf in s^(comment lexbuf))
 733           else comment lexbuf }
 734   (* noteopti: *)
 735   | [^ '*'] { start_line true; let s = tok lexbuf in s ^ comment lexbuf }
 736   | [ '*']   { start_line true; let s = tok lexbuf in s ^ comment lexbuf }
 737   | _
 738       { start_line true; let s = tok lexbuf in
 739         Common.pr2 ("LEXER: unrecognised symbol in comment:"^s);
 740         s ^ comment lexbuf
 741       }
 742